Fun Friday: Share Your Backup Pet Peeves


Userlevel 7
Badge +8

Happy Friday everyone!

 

Today, lets talk pet peeves. If you’re not familiar with the term, it means something of frequent/continual annoyance to someone.

 

Mine is when backup repositories are used for ANYTHING other than backups/data restoration.

 

I see it a lot with people’s personal backups and in smaller businesses. It’s often a scenario such as:

 

I find a file share on a backup repository, and I ask, what’s that doing there? Then I’ll be told that there were some files that were far too big to sit on their expensive storage such as marketing media shares full of videos and photos.

 

I was always taught to assume positive intent so I’ll ask something like “oh okay, have they copied what they need so it can be deleted?”

 

More often than not I’ll be told, nope they need the data still as XYZ department needs it. I’ll then ask the key question, “okay, where’s be being backed up to? And what frequency?” At this point I always get the same response “we have nowhere big enough to back this up to”.

 

The person I’m speaking to normally knows at this point, the point that I’m making, and they can see the problem for themselves. Data on a backup server isn’t a backup if it’s the only copy. At that point you’ve got a live server you’re backing up to, and now you have more problems.

 

That’s my pet peeve off my chest ready to go into the weekend more relaxed, what’s yours?


19 comments

Userlevel 7
Badge +8

I am with you on dumping files on repo servers but also not following simple documentation on how to configure things. We set up jobs, etc. the same way and then you see a cowboy job that follows no standards. 🤣

Userlevel 7
Badge +7

Or people demanding super high precisely specifications for everything, even for things where it is not necessary other which not matters.

But the answers or specifications from the same people are very common and you have to guess what they want… 😎

Userlevel 7
Badge +8

Or people demanding super high precisely specifications for everything, even for things where it is not necessary other which not matters.

But the answers or specifications from the same people are very common and you have to guess what they want… 😎

They’re telling you the answer they need, you have to guess the question & scenarios that get you there 😆 (and they never match!!!)

Userlevel 7
Badge +4

Someone connecting a NAS per iSCSI to the VBR server over a WAN Connection. Most likely with a bandwidth of 10-20 mbit/s. How can this even work :D

This situation seems to follow me over the years :D

Userlevel 7
Badge +8

Someone connecting a NAS per iSCSI to the VBR server over a WAN Connection. Most likely with a bandwidth of 10-20 mbit/s. How can this even work :D

This situation seems to follow me over the years :D

That one hurt!

Userlevel 7
Badge +6

Creating backups of databases via some unmonitored scheduled task instead of relying on Veeam; often the database owner or partner rather wants to create backups in such a way. But the IT department usually can't monitor those backups and still needs to integrate them in the backup concept. When you talk about DR then you notice that the OS and the application itself has been forgotten...and probably the DB guy isn't around to restore his backup.

Someone connecting a NAS per iSCSI to the VBR server over a WAN Connection. Most likely with a bandwidth of 10-20 mbit/s. How can this even work :D

This situation seems to follow me over the years :D

What's wrong with that? It's called internetSCSI, not lanSCSI 🤣

Userlevel 7
Badge +8

Oh boy @regnor, yep that one is horrible. The mental gymnastics I’ve encountered to justify this is insane.

 

DB teams telling me they manage it as they can’t guarantee a member of the backup team will be available when they need to recover. Even though that’s a management issue, not a technical reason.

I also explain in an actual DR scenario, they’d need to rely on a backup member of staff to restore the OS & applications for them to manually restore the data too, making their RTO worse.

I remember being told they had scheduled backups of ALL their databases to a network share, at X time at night. So I asked what time is the network share backed up? Nobody knew! I then had to explain if your share is being backed up earlier in the evening than your databases, you haven’t got any additional copies for nearly a day!

 

Great share @regnor thank you!

Userlevel 7
Badge +7

I have experienced this primary with Oracle DBAs. Some of them have understood the RMAN plugin in the meantime and that they have full control about the db backups. But more of them are very reluctant about this and prefer their local RMAN (!) backups and to backup the files only.

Userlevel 7
Badge +6

This absolutely sounds familiar @MicoolPaul. Everyone wants to stay independent but misses that one alone can only recover 50%. And also it's a bit the fear if something new. Many times its only about showing the DB team what Veeam is capable of and then you'll be able to implement a change.

Userlevel 7
Badge +4

What's wrong with that? It's called internetSCSI, not lanSCSI 🤣


@regnor 
Right, that must be it. I like internetSCSI 😂

For me, when people starts connecting iSCSI units to virtual machines, instead of adding datastores to the virtual environment and adding virtual disks to the vms.

 

Userlevel 7
Badge +4

For me, when people starts connecting iSCSI units to virtual machines, instead of adding datastores to the virtual environment and adding virtual disks to the vms.

 

Depends. If it‘s the iSCSI LUN for the backup repository, I wouldn‘t add an additional layer between the physical disks and the file system.

If my backup repositories with the veeam backup files were on a vsphere datastore which is connected as an iscsi LUN, restoring from a complete disaster takes additional steps. 
First I would have to install a new esxi host to read the content of the vmdk from the vmfs datastore. This could give me a longer RTO.

I prefer to have my backup files not under a dozens layers of different types of storages :-) 

For me, when people starts connecting iSCSI units to virtual machines, instead of adding datastores to the virtual environment and adding virtual disks to the vms.

 

Depends. If it‘s the iSCSI LUN for the backup repository, I wouldn‘t add an additional layer between the physical disks and the file system.

If my backup repositories with the veeam backup files were on a vsphere datastore which is connected as an iscsi LUN, restoring from a complete disaster takes additional steps. 
First I would have to install a new esxi host to read the content of the vmdk with the backup files. This could give me a longer RTO.

I prefer to have my backup files not under a dozens layers of different types of storages :-) 

correct, I mean in a normal use. this usualy happens (in my case) when they are running out of space in a File Server, and have a “old nas” around, and decided to connect it to the file server and continue sharing folders. my usual reaction is 😱 cause I saw tons of “important” folders stored there….

If you want some storage, and you dont care about it, like for tests, isos that you can always re-donwload, it`s ok, but not for adding it to a FS without protecting it.

Userlevel 7
Badge +8

Good one @HunterLF 😆 I’d also expand the old NAS scenario to any old hardware involved in backups. So often you see the hardware that’s been retired due to age, performance and lack of warranty being repurposed for backup/DR. Whenever I see this I look at it and think, so you’ve removed this hardware due to risk, and now you want to use it to mitigate risk? 😬

Good one @HunterLF 😆 I’d also expand the old NAS scenario to any old hardware involved in backups. So often you see the hardware that’s been retired due to age, performance and lack of warranty being repurposed for backup/DR. Whenever I see this I look at it and think, so you’ve removed this hardware due to risk, and now you want to use it to mitigate risk? 😬

🤣right, like backing up in a USB disk, without any other copy, and the disk is like 4 years old, it works, but scares me! a lot!

Userlevel 7
Badge +4

I have 3 pet peeves:

  • Dev workloads, protected accordingly if at all, becoming production
  • Admins who keep bizarre and unnecessary configurations (pRDM, many drives, etc.)
  • Inferior backup targets, unsuitable for their recovery point and recovery time expectations
Userlevel 7
Badge +4

Two more came up…. DFS and EFS.  They skew backup.
DFS is Desktop File System and EFS is Email File System. Users doing this drove me crazy in production and in backup. This is one area where M365 & OneDrive ease things a bit. Now, just getting users not being user-like. 
 

 

Userlevel 7
Badge +8

Two more came up…. DFS and EFS.  They skew backup.
DFS is Desktop File System and EFS is Email File System. Users doing this drove me crazy in production and in backup. This is one area where M365 & OneDrive ease things a bit. Now, just getting users not being user-like. 
 

 

Holy desktop batman!  That would drive me insane!! 🤣

Userlevel 7
Badge +3

Good one @HunterLF 😆 I’d also expand the old NAS scenario to any old hardware involved in backups. So often you see the hardware that’s been retired due to age, performance and lack of warranty being repurposed for backup/DR. Whenever I see this I look at it and think, so you’ve removed this hardware due to risk, and now you want to use it to mitigate risk? 😬

🤣right, like backing up in a USB disk, without any other copy, and the disk is like 4 years old, it works, but scares me! a lot!

Ah yes, this. And when it breaks without warning, I can’t help but think, ‘Told you countless times’

Comment